aarch64: add basic support for sme-f8f16 and sme-f8f32

This patch adds support for the SME_F8F16 and SME_F8F32 features as architecture
options, along with related definitions. This support is required for subsequent
intrinsics to work.

gcc/
	* config/aarch64/aarch64.h:
	(TARGET_STREAMING_SME_F8F16, TARGET_STREAMING_SME_F8F32): Add defines.
	* config/aarch64/aarch64-c.cc:
	(__ARM_FEATURE_SME_F8F16, __ARM_FEATURE_SME_F8F32): Add defines.
	* config/aarch64/aarch64-option-extensions.def:
	(sme-f8f16, sme-f8f32): Add arch options in command line.
	* config/aarch64/aarch64-sve-builtins-functions.h:
	(sme_2mode_function_t): Pass unspec_for_mfp8 parameter through ctor.
	* config/aarch64/aarch64-sve-builtins-sme.def:
	(DEF_SME_FUNCTION_GS, DEF_SME_FUNCTION): Redefine based on
	DEF_SME_FUNCTION_GS_FPM.
	(DEF_SME_ZA_FUNCTION_GS, DEF_SME_ZA_FUNCTION): Redefine based on
	DEF_SME_ZA_FUNCTION_GS_FPM.
	(AARCH64_FL_SME_F8F16, AARCH64_FL_SME_F8F32): Add new
	REQUIRED_EXTENSIONS sections.
	* config/aarch64/aarch64-sve-builtins.cc:
	(TYPES_za_h_mf8): Add new types.
	(TYPES_za_s_mf8): Likewise.
	(sme_function_groups): Define using DEF_SME_FUNCTION_GS_FPM instead of
	DEF_SME_FUNCTION_GS.
	* doc/invoke.texi: (sme-f8f16, sme-f8f32): Add documentation of option.

gcc/testsuite/
	* gcc.target/aarch64/pragma_cpp_predefs_4.c: Add tests checking that
	sme-f8f16 and sme-f8f32 prefefs are off by default, and checks for
	feature dependencies.
	* lib/target-supports.exp: Add check_effective_target support for
	sme-f8f16 and sme-f8f32.

diff --git a/gcc/config/aarch64/aarch64-c.cc b/gcc/config/aarch64/aarch64-c.cc
index 04cd7c354e569e5c190795eee7a977ce81d95108..dcfc14ae9989038c3b30e67bea68930c0a4bbf6d 100644 (file)
--- a/gcc/config/aarch64/aarch64-c.cc
+++ b/gcc/config/aarch64/aarch64-c.cc
@@ -294,6 +294,10 @@ aarch64_update_cpp_builtins (cpp_reader *pfile)
 
   aarch64_def_or_undef (TARGET_SME, "__ARM_FEATURE_SME", pfile);
   aarch64_def_or_undef (TARGET_SME_I16I64, "__ARM_FEATURE_SME_I16I64", pfile);
+  aarch64_def_or_undef (AARCH64_HAVE_ISA (SME_F8F16),
+                       "__ARM_FEATURE_SME_F8F16", pfile);
+  aarch64_def_or_undef (AARCH64_HAVE_ISA (SME_F8F32),
+                       "__ARM_FEATURE_SME_F8F32", pfile);
   aarch64_def_or_undef (AARCH64_HAVE_ISA (SME_B16B16),
                        "__ARM_FEATURE_SME_B16B16", pfile);
   aarch64_def_or_undef (AARCH64_HAVE_ISA (SME_F16F16),

diff --git a/gcc/config/aarch64/aarch64-option-extensions.def b/gcc/config/aarch64/aarch64-option-extensions.def
index c43802eb71543f40f9861686d2fd24f968a97b5d..c8121baa3fd099fbf917dce76c7486622fb1b407 100644 (file)
--- a/gcc/config/aarch64/aarch64-option-extensions.def
+++ b/gcc/config/aarch64/aarch64-option-extensions.def
@@ -308,6 +308,10 @@ AARCH64_OPT_EXTENSION("lut", LUT, (SIMD), (), (), "lut")
 
 AARCH64_OPT_EXTENSION ("sme-lutv2", SME_LUTv2, (SME2), (), (), "smelutv2")
 
+AARCH64_OPT_EXTENSION ("sme-f8f16", SME_F8F16, (SME2, FP8), (), (), "smef8f16")
+
+AARCH64_OPT_EXTENSION ("sme-f8f32", SME_F8F32, (SME2, FP8), (), (), "smef8f32")
+
 AARCH64_OPT_EXTENSION("cpa", CPA, (), (), (), "")
 
 AARCH64_OPT_EXTENSION("fprcvt", FPRCVT, (FP), (), (), "fprcvt")

diff --git a/gcc/config/aarch64/aarch64-sve-builtins-functions.h b/gcc/config/aarch64/aarch64-sve-builtins-functions.h
index c05946d4ec73fb3c4bbe52033f457f98b3787a8c..f5cf6bfb899bb97c985c9a505842724cbb3c0a33 100644 (file)
--- a/gcc/config/aarch64/aarch64-sve-builtins-functions.h
+++ b/gcc/config/aarch64/aarch64-sve-builtins-functions.h
@@ -462,8 +462,9 @@ public:
   using parent = read_write_za<unspec_based_function_base>;
 
   CONSTEXPR sme_2mode_function_t (int unspec_for_sint, int unspec_for_uint,
-                                 int unspec_for_fp)
-    : parent (unspec_for_sint, unspec_for_uint, unspec_for_fp, -1, 1)
+                                 int unspec_for_fp, int unspec_for_mfp8 = -1)
+    : parent (unspec_for_sint, unspec_for_uint, unspec_for_fp, unspec_for_mfp8,
+             1)
   {}
 
   rtx

diff --git a/gcc/config/aarch64/aarch64-sve-builtins-sme.def b/gcc/config/aarch64/aarch64-sve-builtins-sme.def
index 978a74f438ded156db380d88daacbc4b4d23fce7..c86d5fa730bfec395939373b301af9d1e192cf3c 100644 (file)
--- a/gcc/config/aarch64/aarch64-sve-builtins-sme.def
+++ b/gcc/config/aarch64/aarch64-sve-builtins-sme.def
@@ -17,19 +17,25 @@
    along with GCC; see the file COPYING3.  If not see
    <http://www.gnu.org/licenses/>.  */
 
+
+#ifndef DEF_SME_FUNCTION_GS
+#define DEF_SME_FUNCTION_GS(NAME, SHAPE, TYPES, GROUPS, PREDS) \
+  DEF_SME_FUNCTION_GS_FPM (NAME, SHAPE, TYPES, GROUPS, PREDS, unused)
+#endif
+
 #ifndef DEF_SME_FUNCTION
 #define DEF_SME_FUNCTION(NAME, SHAPE, TYPES, PREDS) \
-  DEF_SME_FUNCTION_GS (NAME, SHAPE, TYPES, none, PREDS)
+  DEF_SME_FUNCTION_GS_FPM (NAME, SHAPE, TYPES, none, PREDS, unused)
 #endif
 
 #ifndef DEF_SME_ZA_FUNCTION_GS
 #define DEF_SME_ZA_FUNCTION_GS(NAME, SHAPE, TYPES, GROUP, PREDS) \
-  DEF_SME_FUNCTION_GS (NAME, SHAPE, TYPES, GROUP, PREDS)
+  DEF_SME_ZA_FUNCTION_GS_FPM (NAME, SHAPE, TYPES, GROUP, PREDS, unused)
 #endif
 
 #ifndef DEF_SME_ZA_FUNCTION
 #define DEF_SME_ZA_FUNCTION(NAME, SHAPE, TYPES, PREDS) \
-  DEF_SME_ZA_FUNCTION_GS (NAME, SHAPE, TYPES, none, PREDS)
+  DEF_SME_ZA_FUNCTION_GS_FPM (NAME, SHAPE, TYPES, none, PREDS, unused)
 #endif
 
 #define REQUIRED_EXTENSIONS streaming_compatible (0)
@@ -259,6 +265,12 @@ DEF_SME_FUNCTION_GS (svwrite_lane_zt, write_lane_zt, all_data,  none, none)
 DEF_SME_FUNCTION_GS (svluti4_zt,      luti4_zt,      b_integer, x4,   none)
 #undef REQUIRED_EXTENSIONS
 
+#define REQUIRED_EXTENSIONS streaming_only (AARCH64_FL_SME_F8F16)
+#undef REQUIRED_EXTENSIONS
+
+#define REQUIRED_EXTENSIONS streaming_only (AARCH64_FL_SME_F8F32)
+#undef REQUIRED_EXTENSIONS
+
 #undef DEF_SME_ZA_FUNCTION
 #undef DEF_SME_ZA_FUNCTION_GS
 #undef DEF_SME_FUNCTION

diff --git a/gcc/config/aarch64/aarch64-sve-builtins.cc b/gcc/config/aarch64/aarch64-sve-builtins.cc
index 03481ee4a7757dea0b147069ab7fb1a89a93f856..505a2445d1fcf6aace23ba8822578680cabdf4cc 100644 (file)
--- a/gcc/config/aarch64/aarch64-sve-builtins.cc
+++ b/gcc/config/aarch64/aarch64-sve-builtins.cc
@@ -664,6 +664,10 @@ CONSTEXPR const group_suffix_info group_suffixes[] = {
   TYPES_za_bhsd_data (S, D), \
   TYPES_reinterpret1 (D, za128)
 
+/* _za16_mf8.  */
+#define TYPES_za_h_mf8(S, D) \
+  D (za16, mf8)
+
 /* _za16_bf16.  */
 #define TYPES_za_h_bfloat(S, D) \
   D (za16, bf16)
@@ -700,6 +704,10 @@ CONSTEXPR const group_suffix_info group_suffixes[] = {
 #define TYPES_za_s_integer(S, D) \
   D (za32, s32), D (za32, u32)
 
+/* _za32_mf8.  */
+#define TYPES_za_s_mf8(S, D) \
+  D (za32, mf8)
+
 /* _za32_f32.  */
 #define TYPES_za_s_float(S, D) \
   D (za32, f32)
@@ -849,6 +857,7 @@ DEF_SVE_TYPES_ARRAY (all_za);
 DEF_SVE_TYPES_ARRAY (d_za);
 DEF_SVE_TYPES_ARRAY (za_bhsd_data);
 DEF_SVE_TYPES_ARRAY (za_all_data);
+DEF_SVE_TYPES_ARRAY (za_h_mf8);
 DEF_SVE_TYPES_ARRAY (za_h_bfloat);
 DEF_SVE_TYPES_ARRAY (za_h_float);
 DEF_SVE_TYPES_ARRAY (za_s_b_signed);
@@ -858,6 +867,7 @@ DEF_SVE_TYPES_ARRAY (za_s_h_integer);
 DEF_SVE_TYPES_ARRAY (za_s_h_data);
 DEF_SVE_TYPES_ARRAY (za_s_unsigned);
 DEF_SVE_TYPES_ARRAY (za_s_integer);
+DEF_SVE_TYPES_ARRAY (za_s_mf8);
 DEF_SVE_TYPES_ARRAY (za_s_float);
 DEF_SVE_TYPES_ARRAY (za_s_data);
 DEF_SVE_TYPES_ARRAY (za_d_h_integer);
@@ -987,15 +997,15 @@ static CONSTEXPR const function_group_info neon_sve_function_groups[] = {
 
 /* A list of all arm_sme.h functions.  */
 static CONSTEXPR const function_group_info sme_function_groups[] = {
-#define DEF_SME_FUNCTION_GS(NAME, SHAPE, TYPES, GROUPS, PREDS) \
+#define DEF_SME_FUNCTION_GS_FPM(NAME, SHAPE, TYPES, GROUPS, PREDS, FPM_MODE) \
   { #NAME, &functions::NAME, &shapes::SHAPE, types_##TYPES, groups_##GROUPS, \
     preds_##PREDS, aarch64_required_extensions::REQUIRED_EXTENSIONS, \
-    FPM_unused },
-#define DEF_SME_ZA_FUNCTION_GS(NAME, SHAPE, TYPES, GROUPS, PREDS) \
+    FPM_##FPM_MODE },
+#define DEF_SME_ZA_FUNCTION_GS_FPM(NAME, SHAPE, TYPES, GROUPS, PREDS, FPM_MODE) \
   { #NAME, &functions::NAME##_za, &shapes::SHAPE, types_##TYPES, \
     groups_##GROUPS, preds_##PREDS, \
     aarch64_required_extensions::REQUIRED_EXTENSIONS \
-      .and_also (AARCH64_FL_ZA_ON), FPM_unused },
+      .and_also (AARCH64_FL_ZA_ON), FPM_##FPM_MODE },
 #include "aarch64-sve-builtins-sme.def"
 };
 

diff --git a/gcc/config/aarch64/aarch64.h b/gcc/config/aarch64/aarch64.h
index e3eb807fb53e7dbd21470e00a16d043937ca77eb..deab503d7461fe287b7e5272ad6b5b82ea2a9915 100644 (file)
--- a/gcc/config/aarch64/aarch64.h
+++ b/gcc/config/aarch64/aarch64.h
@@ -321,6 +321,14 @@ constexpr auto AARCH64_FL_DEFAULT_ISA_MODE ATTRIBUTE_UNUSED
 /* The FEAT_SME_I16I64 extension to SME, enabled through +sme-i16i64.  */
 #define TARGET_SME_I16I64 AARCH64_HAVE_ISA (SME_I16I64)
 
+/* The FEAT_SME_F8F16 extension to SME, enabled through +sme-f8f16.  */
+#define TARGET_STREAMING_SME_F8F16 \
+  (AARCH64_HAVE_ISA (SME_F8F16) && TARGET_STREAMING)
+
+/* The FEAT_SME_F8F32 extension to SME, enabled through +sme-f8f32.  */
+#define TARGET_STREAMING_SME_F8F32 \
+  (AARCH64_HAVE_ISA (SME_F8F32) && TARGET_STREAMING)
+
 /* The FEAT_SME_B16B16 extension to SME, enabled through +sme-b16b16.  */
 #define TARGET_STREAMING_SME_B16B16 \
   (AARCH64_HAVE_ISA (SME_B16B16) && TARGET_STREAMING)

diff --git a/gcc/doc/invoke.texi b/gcc/doc/invoke.texi
index ea2d53beb794c7cb9ac40afb7a88848fa38a4bf9..88a87faefdc5c7d901643de6740094fe2e8a2e5e 100644 (file)
--- a/gcc/doc/invoke.texi
+++ b/gcc/doc/invoke.texi
@@ -23082,6 +23082,12 @@ Enable the FEAT_SME_F64F64 extension to SME.  This also enables SME
 instructions.
 @item sme2
 Enable the Scalable Matrix Extension 2.  This also enables SME instructions.
+@item sme-f8f16
+Enable the FEAT_SME_F8F16 extension to SME.  This also enables SME2 and FP8
+instructions.
+@item sme-f8f32
+Enable the FEAT_SME_F8F32 extension to SME.  This also enables SME2 and FP8
+instructions.
 @item sme-b16b16
 Enable the FEAT_SME_B16B16 extension to SME.  This also enables SME2
 and SVE_B16B16 instructions.

diff --git a/gcc/testsuite/gcc.target/aarch64/pragma_cpp_predefs_4.c b/gcc/testsuite/gcc.target/aarch64/pragma_cpp_predefs_4.c
index 3799fb46df125cd040d0445883626c99f1f04552..284c2a232520517125f92c3f1ba74d3bb8e90059 100644 (file)
--- a/gcc/testsuite/gcc.target/aarch64/pragma_cpp_predefs_4.c
+++ b/gcc/testsuite/gcc.target/aarch64/pragma_cpp_predefs_4.c
@@ -70,6 +70,12 @@
 #ifdef __ARM_FEATURE_SME_I16I64
 #error Foo
 #endif
+#ifdef __ARM_FEATURE_SME_F8F16
+#error Foo
+#endif
+#ifdef __ARM_FEATURE_SME_F8F32
+#error Foo
+#endif
 #ifdef __ARM_FEATURE_SME_B16B16
 #error Foo
 #endif
@@ -105,6 +111,34 @@
 #error Foo
 #endif
 
+#pragma GCC target "+nothing+sve2+sme-f8f16"
+#ifndef __ARM_FEATURE_SME_F8F16
+#error Foo
+#endif
+#ifndef __ARM_FEATURE_SME
+#error Foo
+#endif
+#ifndef __ARM_FEATURE_SME2
+#error Foo
+#endif
+#ifndef __ARM_FEATURE_FP8
+#error Foo
+#endif
+
+#pragma GCC target "+nothing+sve2+sme-f8f32"
+#ifndef __ARM_FEATURE_SME_F8F32
+#error Foo
+#endif
+#ifndef __ARM_FEATURE_SME
+#error Foo
+#endif
+#ifndef __ARM_FEATURE_SME2
+#error Foo
+#endif
+#ifndef __ARM_FEATURE_FP8
+#error Foo
+#endif
+
 #pragma GCC target "+nothing+sve2+sme-f16f16"
 #ifndef __ARM_FEATURE_SME_F16F16
 #error Foo

diff --git a/gcc/testsuite/lib/target-supports.exp b/gcc/testsuite/lib/target-supports.exp
index 4d2d1e939d25a5bee634e77a1faafe8b20acd51f..80d22e374e492cf27c33ee9691278e802846359f 100644 (file)
--- a/gcc/testsuite/lib/target-supports.exp
+++ b/gcc/testsuite/lib/target-supports.exp
@@ -12670,6 +12670,7 @@ set exts {
 # We don't support SME without SVE2, so we'll use armv9 as the base
 # archiecture for SME and the features that require it.
 set exts_sve2 {
+    "sme-f8f16" "sme-f8f32"
     "sme-b16b16" "sme-f16f16" "sme-i16i64" "sme" "sme2" "sme2p1"
     "ssve-fp8dot2" "ssve-fp8dot4" "ssve-fp8fma"
 }